Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

A spatio-temporal self-supervised meta-learning network with dynamic graph learning for traffic flow forecasting.

Qian Qiu1,2,3, Yong Huang1,2,3, Xiaoting Huang4

  • 1School of Traffic and Transportation, Guangxi Transport Vocational And Technical College, Nanning, China.

Plos One
|March 27, 2026
PubMed
Summary

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

[The methylation analysis of EMP3 and PCDH-gamma-A11 gene in human glioma].

Zhonghua wai ke za zhi [Chinese journal of surgery]·2010
Same author

[Determination of inorganic elements in the soil-grass-animal system by sealed microwave digestion ICP-AES].

Guang pu xue yu guang pu fen xi = Guang pu·2010
Same author

Induction of genuine autophagy by cationic lipids in mammalian cells.

Autophagy·2010
Same author

[EZH2 expression in human prostate cancer and its clinicopathologic significance].

Zhonghua nan ke xue = National journal of andrology·2010
Same author

Metal-organic frameworks with exceptionally high methane uptake: where and how is methane stored?

Chemistry (Weinheim an der Bergstrasse, Germany)·2010
Same author

Mutation of the protein kinase A phosphorylation site influences the anti-proliferative activity of mitofusin 2.

Atherosclerosis·2010
This summary is machine-generated.

This study introduces the Spatio-Temporal Self-Supervised Meta-Learning Network (SSML-Net) for accurate traffic flow prediction. SSML-Net enhances spatio-temporal modeling and generalizes across diverse traffic scenarios, outperforming existing methods.

Area of Science:

  • Artificial Intelligence
  • Transportation Engineering
  • Data Science

Background:

  • Accurate traffic flow prediction is crucial for managing urban congestion and optimizing transportation systems.
  • Current spatio-temporal graph neural networks face limitations in capturing dynamic spatial dependencies and generalizing across varied traffic conditions.

Purpose of the Study:

  • To develop a novel network, the Spatio-Temporal Self-Supervised Meta-Learning Network (SSML-Net), for improved traffic flow forecasting.
  • To enhance the modeling of complex spatio-temporal dependencies and improve generalization capabilities in traffic prediction.

Main Methods:

  • Proposed SSML-Net integrates spatial and temporal learners with self-supervised meta-learning.
  • Employs meta-learning-driven feature fusion and gated coupling mechanisms to strengthen spatio-temporal interactions.

Related Experiment Videos

Main Results:

  • SSML-Net demonstrated superior performance compared to traditional statistical, deep temporal, and existing spatio-temporal graph-based models on PeMS datasets.
  • Ablation studies confirmed the effectiveness of core components, and small-sample experiments highlighted robustness and generalization.
  • Cross-domain experiments on METR-LR and Beijing traffic data validated exceptional transfer learning and generalization performance.

Conclusions:

  • SSML-Net provides a robust, adaptive, and high-precision framework for spatio-temporal traffic flow prediction.
  • The model effectively adapts to varying data scales and dynamic urban environments, reducing training costs while improving accuracy.